Skip to main content

1. Publish paper or build product? (with Henrique Maia)

Watch it on Youtube. Or listen on your favortie podcast app.

Show note

Henrique and I chat about how I view publishing academic papers versus producitzing cutting edge research. Research engineers and research scientists are both popular job titles that PhD graduates chase after. What's their difference and similarity?

Full transcript text

00:00.000 --> 00:04.880 So your, your, your question was why do I want to be a research engineer?
00:04.980 --> 00:06.840 Like if I can be a research scientist.
00:07.080 --> 00:07.320 Yeah.
00:07.320 --> 00:08.760 I'm facing a dilemma right now.
00:08.760 --> 00:09.000 Yeah.
00:09.000 --> 00:09.280 Yeah.
00:09.320 --> 00:09.720 Yeah.
00:09.820 --> 00:11.720 And I think that's very common.
00:11.720 --> 00:15.660 Actually, I talked with a few headhunters recently and they, they, they kind of
00:15.680 --> 00:18.440 assumed that I would be really passionate about publishing papers.
00:18.640 --> 00:21.840 So they use that as like enticing me, oh, you should come to us because we
00:21.900 --> 00:25.940 encourage everyone to publish paper, but I'm not, I'm I'm interested in products
00:25.940 --> 00:29.880 driven research, like commercialize my research and they're all look surprised.
00:29.880 --> 00:33.880 And I, I'm surprised that they were surprised because if I want to publish
00:33.880 --> 00:36.160 paper, why would I join industry?
00:36.160 --> 00:41.440 I would go to academia, be a professor, and then work with Google folks to get
00:41.700 --> 00:44.760 GPU I need and that with completely academic freedom.
00:45.280 --> 00:45.880 That's good.
00:46.080 --> 00:48.680 But you and I both know that's not exactly true, right?
00:48.680 --> 00:50.560 Like you, you have to write grants.
00:50.600 --> 00:52.740 You don't necessarily get to work on what you want and then you have to
00:52.740 --> 00:56.680 work on what the grant is doing or, or you, or you're like working with students.
00:56.680 --> 00:59.560 Whereas I feel like an industry, there's this sort of idea of like a
00:59.560 --> 01:02.280 blank check, like you can like work on whatever research you want.
01:02.560 --> 01:06.120 If you have like a good enough team and you have all the resources, you don't
01:06.120 --> 01:11.000 have to beg Google or Adobe to give you money, the money's already there in your
01:11.000 --> 01:11.360 pocket.
01:11.440 --> 01:11.680 No.
01:12.400 --> 01:12.680 Yeah.
01:12.720 --> 01:18.760 I think the convincing party just turns from NSF or NIH or, or DARPA, whatever,
01:18.800 --> 01:20.200 to your company.
01:20.440 --> 01:23.480 So you need to convince the company, Hey, I really want to work on this self driving
01:23.520 --> 01:26.240 car research and the company said sure.
01:26.360 --> 01:27.160 Or no.
01:27.160 --> 01:31.520 And then it's, it's, it's, it's the same.
01:31.520 --> 01:34.520 It's the same struggle with the academia, I would say.
01:34.920 --> 01:40.840 And what's better is that you can join a company which share the same vision as you.
01:41.240 --> 01:44.320 Like for example, if you were to join Adobe, it's, you don't need much
01:44.320 --> 01:46.040 convincing if you want to do like.
01:47.800 --> 01:49.800 Cut a person out of a photo.
01:50.040 --> 01:50.840 That's just what we do.
01:50.840 --> 01:53.920 You don't need to convince anyone is if you can do that, that research very well,
01:53.920 --> 01:57.760 you're guaranteed to be founded as in continue to be hired.
01:58.240 --> 02:01.040 And it will be really hard to come in.
02:01.520 --> 02:03.720 To Adobe saying, I want to build a next smartphone.
02:04.000 --> 02:06.800 As far as I know, I don't, I don't, I don't think that's possible.
02:06.920 --> 02:08.280 Or even the self driving car.
02:08.280 --> 02:10.040 That sounds like that would be kind of a stretch.
02:10.360 --> 02:10.720 Yeah.
02:10.920 --> 02:11.200 Yeah.
02:11.200 --> 02:15.680 So, so I think you're faced with the same convincing thing and yeah.
02:16.000 --> 02:19.080 But you're getting compensated while you're doing it, which is a slight
02:19.080 --> 02:23.800 difference from academia, at least as a student from what I'm used to.
02:24.520 --> 02:24.800 Yeah.
02:24.800 --> 02:25.080 Yeah.
02:25.440 --> 02:26.600 But I guess as a professor, yeah.
02:26.600 --> 02:28.280 You, you also deal with like recruiting.
02:29.240 --> 02:31.200 Whereas like, how is it recruiting interns?
02:31.840 --> 02:32.120 Yeah.
02:32.120 --> 02:35.840 Another difference is that as professors, you do need to spend a lot of time
02:36.040 --> 02:39.040 recruiting because once we recruit student, you have a.
02:39.400 --> 02:43.160 Hopefully a long lasting relationship with a student for five to six to seven
02:43.160 --> 02:47.200 years, you need a PhD and hopefully after the graduate, you can still be a mentor,
02:47.200 --> 02:49.120 mentee or friends like relationship.
02:49.760 --> 02:55.360 And, but for company, it's, it's a bit different because we, we, we, um, one of
02:55.360 --> 02:59.200 the primary way that we do research is through interns, intern, uh, research
02:59.200 --> 03:04.040 interns, and sometimes those internship get extended, which means they get
03:04.080 --> 03:07.200 another contract to, to continue to do the research.
03:07.400 --> 03:09.160 Sometimes we don't extend that.
03:09.160 --> 03:13.760 We just let the students go back to the home university and then do the research.
03:13.760 --> 03:18.080 And in either case, the relationship is not on the scale of years.
03:18.200 --> 03:19.640 It's usually on a scale of months.
03:20.200 --> 03:27.560 And if you, if you happen to hire a intern who share the same research path as you,
03:27.760 --> 03:30.640 you might be able to convince them to come back and do the multiple internship
03:30.640 --> 03:35.440 and then essentially turn that into a pseudo PhD student for you.
03:35.440 --> 03:35.680 Yeah.
03:35.680 --> 03:39.000 Um, But I don't see that happen very often.
03:39.000 --> 03:39.560 It happens.
03:39.560 --> 03:42.680 It does happen within Adobe, within other like research labs.
03:42.680 --> 03:44.240 But it doesn't happen very frequently.
03:44.560 --> 03:48.360 And for my interns, I actually encourage them to intern elsewhere after
03:48.360 --> 03:52.040 interning once or twice with me, because although I think I'm the best mentor out
03:52.040 --> 03:56.800 there, but they should see all the other companies, the other mentors, and then
03:56.800 --> 04:00.360 find out which style that they want and which style they want to grow into.
04:00.640 --> 04:03.680 So, so far, I haven't had a three-time intern.
04:03.680 --> 04:07.800 I have a couple of two-time interns and after two times I said, you should
04:07.800 --> 04:09.400 really should find another place.
04:09.880 --> 04:11.720 And I really enjoy working with all of them.
04:11.720 --> 04:13.760 Yeah, that's, it's like lower stakes.
04:13.760 --> 04:17.960 Like if you choose a PhD mentor, you're stuck with them for five, six years,
04:17.960 --> 04:19.280 and hopefully it's like a really good relationship.
04:20.000 --> 04:23.280 But then when you're, you're testing, you can test out lots of different industries,
04:23.280 --> 04:27.640 lots of different companies and see how they all feel if you go that route.
04:27.760 --> 04:32.160 But I guess from, from your perspective, grabbing students to work with, you're
04:32.160 --> 04:36.040 only working on small scale projects or like they're moving along a project one
04:36.040 --> 04:39.080 step forward for someone else to pick it up and eventually finish it.
04:39.600 --> 04:39.760 Yeah.
04:39.960 --> 04:40.280 Yeah.
04:40.280 --> 04:42.840 Yeah, yeah, it's like, you're right.
04:42.840 --> 04:48.080 Like it's smaller scope and yeah, you as the man, as the industry full-time
04:48.080 --> 04:51.720 researcher, you need to somehow stitch them together and yeah.
04:52.520 --> 04:52.800 Yeah.
04:53.280 --> 04:57.440 I mean, so then research engineering, now you get to just your products
04:57.440 --> 05:01.840 actually make it to like an audience, like the, to the people, you know, you
05:01.840 --> 05:03.960 can tell them about what you're working on and they can actually experience it.
05:04.400 --> 05:07.600 But you don't necessarily have the choice or as much freedom as you have as a
05:07.600 --> 05:11.280 scientist, or what's the difference there, but building a product usually
05:11.280 --> 05:13.880 has more constraints because you're constrained by a lot of realities.
05:13.880 --> 05:18.040 Like with papers, you only need to convince three reviewers and you can
05:18.040 --> 05:23.280 try multiple times, but with products, you only got a few trials before
05:23.280 --> 05:26.240 company gives up on certain, certain attempts.
05:26.760 --> 05:31.400 And you also need to work with usually, hopefully a large team of other people,
05:31.440 --> 05:34.120 researcher engineers, like designers, you name it.
05:34.120 --> 05:38.560 And, and so that comes with a lot of trade-offs.
05:38.560 --> 05:41.240 Like you can't work on whatever that you think is the best.
05:41.240 --> 05:45.160 You want to align with the team and team sometimes will, it depends on team.
05:45.160 --> 05:46.840 Sometimes a bottom ups and I top down.
05:47.080 --> 05:51.560 Basically each one will have their, their ownership of the work and
05:51.560 --> 05:53.800 they just need to push forward in that particular ownership.
05:53.800 --> 05:56.720 You cannot just say, I do not want to work on this area anymore.
05:56.960 --> 05:58.080 I'm switching to other parts.
05:58.280 --> 06:00.960 If you do that, then people won't trust you anymore.
06:01.400 --> 06:01.560 Yeah.
06:01.560 --> 06:02.520 So you won't be part of it.
06:02.800 --> 06:03.240 Right.
06:03.240 --> 06:03.640 Right.
06:04.040 --> 06:04.280 Right.
06:04.280 --> 06:09.680 And the, the bad, the advantage of building such like product types
06:09.680 --> 06:14.320 research as I call it is, is you get amplified, like, for example, I'm
06:14.320 --> 06:18.480 only working on a part of a big project.
06:18.480 --> 06:21.200 For example, I work on video understanding, for example, I
06:21.200 --> 06:23.600 generate video tags for videos.
06:24.160 --> 06:28.360 And, but I don't know any other things about like shipping a product.
06:28.360 --> 06:29.760 I don't know how to do logins.
06:29.800 --> 06:31.840 I don't know how to handle concurrency.
06:31.840 --> 06:35.680 I don't know how to do Kubernetes, you know, all this fancy words that I don't
06:35.680 --> 06:42.080 know, and that's when my impact get amplified without them, my, my, my
06:42.120 --> 06:48.480 research will only be a Google club notebook and nobody will actually use it.
06:48.520 --> 06:48.720 Yeah.
06:48.720 --> 06:52.440 It won't suit their workflow, but by working in this bigger team, hopefully
06:52.440 --> 06:55.880 we will think about what's the workflow, what's the user journey, and then how can
06:55.880 --> 06:57.360 we help them achieve their task.
06:57.360 --> 07:01.840 I think that's, that's, that's helping me to kind of prioritize my work.
07:02.480 --> 07:05.960 So as you've kind of like shifted from research scientist to engineer, it
07:05.960 --> 07:08.280 sounds like have you been working with larger teams?
07:08.280 --> 07:11.640 Like, I guess as a scientist, it's like you, maybe one other collaborator
07:11.840 --> 07:15.520 or, and then like some students versus like a research engineer, you're part
07:15.520 --> 07:16.960 of this, like a group working on a project.
07:17.360 --> 07:17.720 Yeah.
07:17.760 --> 07:22.120 Like, yes, as like, I think the title of research scientist and research
07:22.120 --> 07:23.880 engineer is sometimes misleading.
07:23.880 --> 07:26.360 So I will use, uh, the work context.
07:26.360 --> 07:30.280 So when I'm working on the research paper with someone is usually a small cohort,
07:30.560 --> 07:34.960 sometimes as small as two, me and my, my, uh, intern and.
07:35.480 --> 07:39.240 And the biggest time, maybe five, maybe with a few other, we researcher and
07:39.240 --> 07:41.360 probably their university advisor.
07:41.400 --> 07:44.920 That's as much as you get like five, but with this kind of product
07:44.920 --> 07:49.560 highs prototype thingy, uh, there's no, like the lower bound is probably
07:49.560 --> 07:52.440 five to 10 people because you need certain people to get to gain momentum.
07:52.840 --> 07:54.680 And then there's really no upper limit.
07:54.680 --> 07:59.120 I, yeah, like you can, as long as you can convince people to join your
07:59.120 --> 08:02.080 efforts, you can, you can, yeah, you can group all of them.
08:02.800 --> 08:06.840 And then, and how long is that sort of project last, like the internship
08:06.840 --> 08:09.960 projects last couple months, and then maybe you publish a couple papers
08:09.960 --> 08:11.400 a year or one paper a year.
08:11.920 --> 08:12.160 Yeah.
08:12.160 --> 08:16.720 For internship, it's Adobe's intern program is, um, 13 weeks plus the July
08:16.720 --> 08:17.480 4th shutdown week.
08:17.480 --> 08:19.120 So calendar week is 14 weeks.
08:19.720 --> 08:21.080 Um, yeah.
08:21.120 --> 08:23.880 And then afterwards, either we extend the internship.
08:23.880 --> 08:26.960 Or we just let the student go back to the university, taking
08:26.960 --> 08:28.200 their code with them, of course.
08:28.560 --> 08:30.840 And then they, they just keep working on it.
08:31.040 --> 08:35.040 And usually, um, for a different project, you had different time spans,
08:35.520 --> 08:39.680 like a project in graphics, like SIGGRAPH usually require more
08:39.680 --> 08:43.680 implementation, more system building tends to get longer a year, maybe longer.
08:44.000 --> 08:47.720 And a computer vision, an HCI project tend to be on the shorter side, like
08:47.720 --> 08:50.120 maybe half a year, a year will be enough.
08:50.160 --> 08:53.320 It depends heavily on the project, the students, the fit, the
08:53.320 --> 08:54.360 fits and everything.
08:54.520 --> 08:54.760 Yeah.
08:55.520 --> 08:59.320 But then the research engineers are the more like prototype
08:59.320 --> 09:00.480 product projects.
09:00.520 --> 09:04.200 Those are the ones they can take up anywhere from like a year to two years.
09:05.040 --> 09:05.320 Yeah.
09:05.880 --> 09:06.280 Yeah.
09:06.360 --> 09:06.640 Yeah.
09:06.640 --> 09:11.600 They can, they can decide how long to stick to a particular efforts.
09:12.040 --> 09:15.520 I mean, as long as the effort is still going, if not, I feel by like,
09:15.680 --> 09:16.920 yeah, higher level.
09:17.240 --> 09:17.480 Yeah.
09:17.800 --> 09:18.000 Yeah.
09:18.000 --> 09:19.320 And so I guess who decides that?
09:19.320 --> 09:22.240 So it's like the higher ups, like your manager or your manager's manager,
09:22.240 --> 09:23.800 is the one kind of leading that effort.
09:24.360 --> 09:24.720 Yeah.
09:24.760 --> 09:28.760 I think, I think I, in my personal experience, there's not a, like a
09:28.760 --> 09:32.040 mandate, Hey, Dean, you have to work on this particular thing.
09:32.440 --> 09:36.680 It's more like I just grab, grab all the information that can have.
09:36.760 --> 09:42.240 And then I make my own judgment on which team or which tech transfer
09:42.240 --> 09:43.800 or which research I should work on.
09:44.360 --> 09:49.160 It's and I think working on this particular efforts, because I trusted
09:49.160 --> 09:52.040 in this effort, I think it will have good return in a few years.
09:52.040 --> 09:52.880 It's like investment, right?
09:52.880 --> 09:56.840 Like some people choose this stock, that stock I choose and like
09:56.840 --> 09:58.040 an index fund or whatever.
09:58.320 --> 10:00.280 So that, that, that, that suits me.
10:00.640 --> 10:04.480 And similarly here, like usually there are lots of efforts going on, either
10:04.720 --> 10:09.400 research paper efforts or like research prototype, like future product efforts.
10:10.040 --> 10:15.840 And each researcher will just use their own judgment to align their
10:15.840 --> 10:17.040 interests with what they like.
10:17.080 --> 10:17.360 Right.
10:17.520 --> 10:17.720 Yeah.
10:17.720 --> 10:18.000 Yeah.
10:18.000 --> 10:22.320 So like from the inside, like you guys are all there, you see someone
10:22.320 --> 10:25.640 who is like working on just like projects and prototypes and other
10:25.640 --> 10:29.800 people working on papers, everyone kind of gets treated the same from the
10:29.800 --> 10:33.440 inside, like there's no sort of like two different groups or judgments.
10:33.440 --> 10:37.040 It's all mixed around and people kind of waved back and forth between the two.
10:37.360 --> 10:37.640 Yeah.
10:37.680 --> 10:37.880 Yeah.
10:37.920 --> 10:42.640 People are the same, like there's no, yeah, like it's, it's very fluid.
10:42.640 --> 10:46.200 Like I can still, I still publish just not, I just don't spend a hundred
10:46.200 --> 10:48.880 percent of my time mentoring interns anymore.
10:48.880 --> 10:52.560 Sometimes I work with university professors directly as a collaboration
10:53.160 --> 10:57.600 and, and then sometimes people spend all their time working on papers.
10:57.640 --> 11:03.120 I think it's, it's a, it's a continuous slider and yeah, we are in the same
11:03.120 --> 11:05.960 group and no, there's, I don't think there is any different treatment.
11:06.880 --> 11:10.200 And I guess, yeah, what's the incentive then it's just whatever drives you.
11:10.200 --> 11:14.840 If you want to publish more papers that sort of builds up your, your academic
11:14.840 --> 11:18.920 profile, or if you just want to keep working on prototypes or you're like
11:19.000 --> 11:24.080 fascinated by a certain project, or there's also like financial, I don't know.
11:24.920 --> 11:25.240 Yeah.
11:25.280 --> 11:30.080 I think, I think our industry, one of the incentives is the, the kind of
11:30.080 --> 11:35.040 the rewards every year, and then that usually it's measured by this intangible
11:35.040 --> 11:39.480 word of impact and it's, it's a, it consists of many, many aspects.
11:39.480 --> 11:41.400 It could be papers, it could be tech transfers.
11:41.400 --> 11:42.240 It could be new product.
11:42.240 --> 11:43.680 It could be a lot of things.
11:43.680 --> 11:45.680 So then that's also brings back to the topic.
11:45.680 --> 11:48.840 There's no different treatment because everything can be counted as an impact.
11:49.360 --> 11:54.360 And, and, and it depends on whether the thing you're working on is, is impactful.
11:54.600 --> 11:58.200 Like you're publishing a lot of, uh, paper that's not really
11:58.200 --> 11:59.440 insightful, it's not helpful.
11:59.800 --> 12:03.320 And if you publish one like best paper that lots of people are talking about
12:03.320 --> 12:05.680 and inspiring, that's really helpful.
12:05.800 --> 12:09.360 Similarly with, with product, you can ship tiny features like that.
12:09.360 --> 12:13.320 Nobody's going to use every quarter or you can wait a few years and ship
12:13.320 --> 12:16.560 a new Photoshop that would be much more impactful.
12:17.040 --> 12:22.880 Um, so I think, yeah, I guess it both helps in her that it's like intangible.
12:22.880 --> 12:24.880 It's just like this nebulous impact.
12:25.200 --> 12:25.880 Exactly.
12:26.120 --> 12:27.360 Nebulous is the right term.
12:27.400 --> 12:27.640 Yeah.
12:27.680 --> 12:28.040 Yeah.
12:29.240 --> 12:30.120 Well, that's exciting though.
12:30.200 --> 12:32.240 Um, yeah.
12:32.480 --> 12:35.280 What, what sort of impact are you looking to make in the next couple of years?
12:35.280 --> 12:42.360 Um, I, since I graduated, I, I think I slowly transitioned from wanting to
12:42.360 --> 12:46.480 publish a lot of paper to wanting to build some, uh, like research
12:46.480 --> 12:48.280 prototypes that can help end users.
12:48.760 --> 12:53.440 And that can be seen as my kind of drop in the number of papers that I work on.
12:53.480 --> 12:58.720 Because, because I just no longer value papers that much at my current
12:58.720 --> 13:00.840 stage of career, maybe later on, I will.
13:00.840 --> 13:05.160 Prank up my paper publishing ability, but right now I'm fine with, okay, just
13:05.160 --> 13:07.000 keep it at a maintainable level.
13:07.000 --> 13:11.160 Just keep me in touch with the research community while I spend a lot of my
13:11.160 --> 13:16.760 working time building whatever research prototypes and products that, yeah.
13:17.320 --> 13:20.800 Do you think there was like a moment where like someone picked up one of
13:20.800 --> 13:24.480 your research prototypes and they like, like playing with it that you said,
13:24.480 --> 13:25.840 okay, I want to make more products.
13:26.280 --> 13:28.760 Or was there like a, just a product that you wanted to make?
13:28.760 --> 13:31.280 Like there, there was at least some inflection point, right?
13:31.280 --> 13:33.680 Like where you started going, all right, I'm going to work more on product.
13:34.640 --> 13:39.200 Um, I think, I think, I think it's, it's, it's, uh, it's, um, you know, I think
13:39.200 --> 13:43.320 there is internal and external factors for me.
13:43.320 --> 13:47.440 I think I've been always been like, I'm fascinated by, by, by
13:47.680 --> 13:49.400 applications, product users.
13:49.520 --> 13:54.080 I'm not so much, uh, like, Oh, I want to learn all the math and
13:54.080 --> 13:55.600 physics behind all the phenomenon.
13:55.600 --> 13:57.640 I'm just fascinated by, by, by.
13:57.640 --> 13:58.600 Behind all the phenomenon.
13:58.600 --> 13:59.760 I'm just not that person.
14:00.320 --> 14:06.000 And then also externally, I think being exposed in a, in a industry research lab,
14:06.360 --> 14:11.840 I just get to see a lot of the, the process from a early stage prototype
14:11.840 --> 14:13.840 to how, to the end, that's delivered.
14:14.200 --> 14:15.560 And it was all fascinated by that.
14:15.920 --> 14:20.160 And I thought, Oh, I want to try that and see how it's like, and, and because
14:20.160 --> 14:26.240 what a paper one, um, I published some, not a lot, but maybe a few dozens of
14:26.240 --> 14:31.200 papers and I think I've got the sense of the end to end pipeline from ideation to.
14:31.920 --> 14:35.480 Writing the paper and make, like writing the cover in the paper and then getting
14:35.480 --> 14:37.000 the presentation and the advertising.
14:37.000 --> 14:40.360 I think that pipeline, I have a pretty good handle of what it is,
14:40.680 --> 14:45.200 but for this like new thing, uh, yeah, I'm more excited for now.
14:45.640 --> 14:49.440 It's almost a novelty of like, trying to understand how that pipeline works.
14:49.480 --> 14:52.840 If you ever wanted to make a product, now you would understand a little bit better
14:52.840 --> 14:55.280 about the whole, the whole pipeline ends and outs.
14:55.280 --> 14:55.480 Yeah.
14:55.480 --> 14:58.520 And also another frustration that I have is a lot of my own research,
14:58.760 --> 15:05.240 not to criticize anybody else's, my own research, most of them are not directly
15:05.560 --> 15:08.280 applicable to any products, which is understandable.
15:08.280 --> 15:11.360 We are aiming like five, 10 years, like 20 years ahead.
15:11.800 --> 15:16.520 And, and, but again, I want to work on product driven research.
15:16.760 --> 15:19.640 So I want to be able to kind of build this prototype to understand, to better
15:19.640 --> 15:24.200 understand, Hey, if I were to work on two intern project next year, which topic
15:24.200 --> 15:28.680 I should choose so that they are more grounded so that they are more impactful
15:28.720 --> 15:32.720 in the sense that it can directly impact existing product or new products.
15:33.000 --> 15:35.960 I don't have to wait five, 10 years before that happens.
15:36.360 --> 15:36.640 Yeah.
15:36.760 --> 15:40.240 It's less a theory or an idea of like, Oh, if we had all the technology,
15:40.240 --> 15:44.360 this would be great, but more of like, what can we actually do that's within
15:44.360 --> 15:46.720 reach now, but isn't being done right now.
15:47.440 --> 15:47.720 Yeah.
15:47.720 --> 15:48.000 Right.
15:48.240 --> 15:48.560 Right.
15:48.560 --> 15:54.720 Right. Well, that only makes it harder for me to decide what I want to do.
15:55.000 --> 15:58.080 I want to do research engineering or a research scientist.
15:58.680 --> 16:02.320 Uh, I feel like I still need to accumulate a little bit more of like
16:02.320 --> 16:05.200 momentum with papers in terms of like figuring out the pipeline and like
16:05.200 --> 16:10.440 swimming out, like how to, how to work with collaborators, um, like how to
16:10.440 --> 16:14.680 go through the ins and outs of like ideation, exploration research and paper.
16:15.280 --> 16:15.640 Yeah.
16:15.720 --> 16:15.880 Yeah.
16:15.880 --> 16:17.080 I think, I think you made a great point.
16:17.080 --> 16:22.240 You, you, you, you wanted to say you want a few more papers in the pipeline so
16:22.240 --> 16:24.240 that you can get to know all the ins and outs.
16:24.680 --> 16:28.080 I think I, I don't know if I talked this with you or somebody else.
16:28.520 --> 16:32.520 I definitely share this with a lot of my interns, which is I encourage them to
16:32.920 --> 16:37.120 value quantity versus quality at the beginning of their PhD.
16:37.120 --> 16:40.800 So try to publish more, don't worry too much about a quality.
16:40.800 --> 16:41.920 Of course you don't want to publish.
16:41.920 --> 16:47.200 Yeah, but, but still you want to focus on the quantity because once you get
16:47.200 --> 16:50.840 the pipeline running, then you know which part you're lacking, which part you enjoy.
16:51.240 --> 16:54.400 Maybe very quickly, you realize you really hate writing codes.
16:54.720 --> 16:58.040 Then you should try to pivot project that does more, I don't know, math,
16:58.080 --> 17:03.840 more physics equations, and maybe you really like 3d printing, fixing 3d printers.
17:03.840 --> 17:07.720 Then you should gravitate towards fabrication, how to improve existing
17:07.720 --> 17:12.280 hardware and, but without working on many projects, many means probably three to
17:12.280 --> 17:17.320 five more, like then without working on that many, like you won't have that sense.
17:17.840 --> 17:18.720 And yeah.
17:19.040 --> 17:19.440 Yeah.
17:19.440 --> 17:24.120 Even just like surviving a deadline, you're each deadline gets progressively easier.
17:24.440 --> 17:27.160 There there's always hiccups along the road, or you're not really
17:27.160 --> 17:28.000 sure what's going to happen, but.
17:28.800 --> 17:31.160 You just start to get a sense for like, what's more important
17:31.160 --> 17:32.120 to be spending your time on.
17:32.720 --> 17:36.360 And then like, once you free up the time that you're wasting, you
17:36.360 --> 17:39.920 have the time that you were wasting before just like running cycles or
17:39.920 --> 17:43.120 running experiments that weren't going to be used, then you start becoming
17:43.120 --> 17:46.080 more efficient and at least better at paper writing.
17:46.680 --> 17:48.200 That's, that's what I found myself also.
17:48.400 --> 17:48.680 Yeah.
17:48.760 --> 17:49.720 Figure out that story.
17:50.280 --> 17:50.560 Yeah.
17:50.720 --> 17:51.000 Yeah.
17:52.960 --> 17:57.400 Um, but yeah, so now, now I guess let's turn to the personal stuff.
17:57.720 --> 17:58.080 Okay.
17:58.080 --> 18:08.080 Let me stop the recording.